Dynamic Causal Effect Evaluation Method Research And Application

The causal inference method based on the theory of randomized experimental design is a revolution of epistemology and methodology,which provides a new cognitive perspective and research method for the evaluation of policy effects in social sciences.The counterfactual analysis method based on the framework of potential results makes up for the shortcomings of traditional causal inference research,which is conducive to reveal the causality between economic variables and improve the scientific level of policy effect evaluation.However,the early causal effect analysis methods mainly focus on the evaluation of microeconomic policies or social projects based on cross-sectional data or short(micro)panel data.Restricted by the data structure,the average processing effect of policy intervention only reflects the static causal effect of economic policies.Because of the sequence correlation of macroeconomic variables,the policy intervention in a certain period often has a continuous impact on the current period and the lag period of the outcome variables.Therefore,the static analysis method of policy causal effect cannot be used directly to measure and reflect the dynamic causal effect of macroeconomic policies,so it is urgent to develop and improve the theory of dynamic causal effect analysis of macroeconomic policies.The identification and estimation of dynamic causal effect is one of the core problems in Macroeconometrics,which is of great significance to evaluate the causal effect among variables and to mine the causal relationship among multiple variables.At present,the existing dynamic causal inference methods based on single equation time series include interruption time series method,dynamic matching method,sequential matching method and inverse probability weighting method.The dynamic causal inference method for multiple time series system is mainly based on the structural econometric model under certain recognition constraints to investigate the impact of policy shocks on the economic system Most of the existing researches are based on SVAR model and DSGE model.In fact,when evaluating the macroeconomic policy effect of a country,because of its special economic structure and political system,it is difficult to find the matching control group economy,which limits the application of causal effect analysis methods such as composite control.In addition,in view of the long-term sustainability of macroeconomic behavior,it is necessary to identify the long-term trend and short-term volatility of the dynamic effects of economic policies.Therefore,it is necessary to establish and improve the dynamic causal effect analysis method based on non-stationary time series model.Therefore,based on the systematic review of domestic and foreign causal effect assessment methods and empirical research,this paper improves the dynamic causal effect analysis method based on time series data from the following three aspects,and makes an empirical analysis of China's economic problems.First,in order to avoid the evaluation bias of the causal effect caused by the heterogeneity of policy intervention samples and the "pseudo regression" of the interruption time series method in the synthetic control method,as well as to relax the limitation of the "second-order moment stationary" of the result variable,and realize the estimation and inference of the dynamic causal effect by using the time series data,this paper uses the cointegration theory of the time series,based on the error correction model and the junction A dynamic causal effect inference method of cointegration time series is proposed by construction mutation test.This method can not only infer the existence of policy intervention effect,but also effectively identify the long-term effect and short-term fluctuation effect of policy intervention on outcome variables.In addition,the paper studies the causality of China's real estate tax pilot policies by using the dynamic causality inference method of cointegration time series.Taking Shanghai and Chongqing as.examples,this paper studies the long-term and short-term effects of the real estate tax policy on the price of commercial housing in 2011.It is found that the real estate tax pilot policy has a significant long-term effect on restraining the price rise of commercial housing;and the policy effect of the two cities shows certain differences due to the system design.The demand squeezed out by the real estate tax pilot policy implemented in nine District of Chongqing city significantly pushes up the price of small-sized housing,produces structural effect,making the policy strength smaller and the time lag longer,while Shanghai City The pilot policy of real estate tax implemented in the whole region is powerful and effective in restraining the rise of house prices.Second,in order to improve the ability and efficiency of the dynamic causal effect of economic policy on macroeconomic regulation,the asymmetry of demand side economic policy effect in different regional systems is studied.This paper states the dynamic causal effect from the perspective of potential results analysis framework,and discusses the identification constraints equivalent to the conditional independence hypothesis and the inverse probability weighted estimation of dynamic causal effect Method.According to people's study and judgment of economic development process,a country's economy is defined as different regional systems(system,or periodic stage)by using some structural mutation test,and the long-term trend factors of each regional system are eliminated by using H-P filtering,BP filtering and regression methods,and the short-term fluctuation variables of economic system in each regional system are obtained,which are defined as responding to the demand side economic policy shock Sequence after.Then,we deduce the dynamic effect of economic policy by establishing single equation ECM model or SVAR/SVEC model.Finally,we define the long-term "growth" and "recession" regime of macro-economy based on catastrophe theory,and take the benchmark interest rate policy of deposit and loan as an example,we propose the asymmetry of dynamic causal effect of different regimes and different directions of economic(monetary)policy shocks by constructing Wald statistics Statistical test method.Moreover,this method can be extended to solve the asymmetric inference problem of the average causal effect of other two zone general policies.Thirdly,in order to solve the modeling problem of non-stationary variables and identify the long-term and short-term dynamic causal effects between variables,this paper uses the simplified VMA/SVMA model of SVAR model,VEC model and SVEC model to carry out the theoretical and application research on the inference of dynamic causal effects of macroeconomic system,and puts forward the dynamic causal effect analysis of non-stationary(variable)economic system policy based on SVEC-? Methods,the long-term random trend(causal)effect and short-term dynamic causal effect of interest rate shocks on endogenous variables were decomposed and identified according to the SVMA model.Finally,the paper uses the SVEC-? method to study the dynamic causal effect of China's monetary policy,identifies the effectiveness of benchmark interest rate policy,and studies the heterogeneity of the impact of interest rate policy on the long-term stochastic trend of non-stationary macro variables such as industrial added value and CPI,as well as the short-term dynamic causal effect.